B-Spline-based Torque Estimation Algorithm Using Dual Encoders

The consideration of joint torques enables collaborative robots to perform various tasks. Recent advancements in encoder resolution have enabled measuring joint torque based on dual encoders mounted on the motor side and the load side. The joint torque was estimated using the models of a kinematic error and the hysteresis and torsional stiffness of a harmonic drive. Admittedly, various models have been adopted for the hysteresis and torsional stiffness. However, only a simple model based on the Fourier series has been proposed for the kinematic error, which does not sufficiently compensate for the kinematic errors owing to various reasons. This study proposes a novel method based on B-spline to compensate for kinematic errors. The proposed method generates waypoints that allow the B-spline to accurately compensate for kinematic errors. Moreover, the proposed method was experimentally verified and compared with an existing model based on the Fourier series. Evidently, the experimental results of the proposed method present a more accurate torque estimation.